Sie sind auf Seite 1von 11

Fundamentals of Transformer Inrush

Suhag P. Patel, P.E.

microprocessor relays, this is done by mathematical


Abstract — This paper will present an overview on the calculation. Phase compensation in electromechanical relays
phenomenon of transformer inrush and its impact on is performed externally by CT connections, and again, by
differential relays. In order for the protection engineer to mathematical calculation in microprocessor relays. Figure 1
properly address the phenomenon, the theoretical nature of shows a simplified diagram to explain how an
inrush will be discussed. Impact of residual flux and higher
electromechanical transformer differential relay is connected.
efficiency core designs will be covered. Examples of inrush
waveforms will be analyzed in the time domain as well as in the
frequency domain.
Modern protective relaying techniques to detect and prevent
misoperation will also be covered. Design of transformer
differential inrush restraint methods including: second
harmonic, fourth harmonic, total harmonic, per phase, cross
blocking, adaptive restraint and other methods will be covered
as well. Discussion on the impact of motor/generator differential
relays due to inrush will also be covered. Techniques for
avoiding nuisance tripping of machine differential relays
without compromising speed of operation will be discussed.

I. INTRODUCTION
For most power transformers with a rating of 10 MVA or
higher, differential relaying is the primary means of
transformer protection. The differential principle as applied
to a transformer is very simple to understand, the sum of all
winding currents, corrected for magnitude and phase shift, Figure 1: Simplified Electromechanical Differential Relay.
should be zero during normal operation. If the sum of the
currents are non-zero, a fault is present. In a perfect world During high magnitude external faults it is possible one or
this idea works very well, however, because of current more winding current transformers may saturate. This creates
transformer (CT) accuracy, CT saturation, tap changer a problem for the differential relay because the secondary
operation, transformer losses, and inrush current the sum of currents will no longer appear equal and can cause
all winding currents is not always zero. This paper will focus misoperation as shown in Figure 2. The most common
on the transformer inrush phenomenon, how it impacts method to address the saturation effect is to use some type of
differential relaying, and techniques applied in various percent differential characteristic, such as the dual percentage
transformer differential relays to address the inrush slope characteristic shown in Figure 3. With this type of
phenomenon. characteristic the amount of differential current required to
Additionally, transformer inrush can have impacts beyond trip is increased as the measured winding current increases.
the transformer differential relay, particularly in the case of a This is an intuitive concept, because as total current
small generator energizing a transformer. In this case careful increases, the probability of CT saturation also increases.
consideration must be given to the generator differential Transformer differential relays can also misoperate due to
relaying to ensure misoperations do not occur due to transformer inrush, which will be explained in detail during
transformer energization. the next section. During inrush conditions, as shown in
Figure 2, the differential relay measures a high magnitude of
II. TRANSFORMER DIFFERENTIAL RELAYING PRINCIPLES current on one winding and measures very little or no current
A transformer differential relay is able to take multiple on the other winding. This clearly meets the definition of a
winding currents and put them on a per unit basis for fault condition, however, we know that this is a normal
comparison. Magnitude compensation in electromechanical condition during transformer energization. A method to
relays is done through the use of tap settings, and in detect this condition and block operation of the differential
relay must be employed to prevent misoperation under these
conditions. Various methods to detect this inrush condition
S. P. Patel is with General Electric Inc., Placentia, CA 92871 USA exist and will be discussed in section four of this paper.

978-1-4577-0496-3/11/$26.00 ©2011 IEEE 290 Protective Relay 2011


,PDJQHWL]LQJ =  − η • )/&
ZKHUHη =WUDQVIRUPHUHIILFLHQF\
)/& = )XOO /RDG &XUUHQW

As most transformers are very efficient under steady state


conditions, very little magnetizing current flows, typically one
to five percent of full load rating. As shown in Figure 4, this
current causes differential current to flow. However, during
steady state conditions, this low magnitude error is easily
accounted for by proper engineering of the differential relay
pickup settings.

5 ;

, ,

Figure 2: Differential Relay Under CT Saturation and Inrush.


0DJQHWL]LQJ
&XUUHQW 5P ;P

Figure 4: 1:1 Transformer Equivalent Model.

The waveform of this steady state magnetizing current is


highly non-linear. The non-linearity is due to the
characteristics of the electrical steel used in the core.
Electrical steel has a high magnetic permeability, but it is not
constant, creating linear and saturation regions of operation.
These characteristics combine to form the familiar hysteresis
curve shown in Figure 5. Figure 5 also shows the relationship
between excitation voltage, flux, and excitation current for a
Figure 3: Dual Slope Percentage Differential Characteristic.
transformer that has been energized and is in steady state
operation.
III. TRANSFORMER INRUSH PHENOMENON
ϕ
H ϕ ϕ

Anytime the excitation voltage applied to a transformer is


ϕ
ϕ

changed, a magnetizing inrush current flows. Although


magnetizing inrush is typically considered to occur when a ϕ

de-energized transformer is energized, magnetizing inrush Time ϕ


Lϕ Lϕ
ϕ


can also flow after system voltage dips and during post fault Lϕ
voltage recovery. This magnetizing inrush current flows
because all transformers must establish flux in the core to
magnetically link the windings together. The flux is
established anytime an exciting voltage is applied to the Figure 5: Transformer Core Excitation Phenomena
transformer. The flux in the core requires a magnetizing
current to maintain itself, as shown in the transformer Although steady state magnetization currents cause small
equivalent circuit of Figure 4. As a rule of thumb, this steady errors in differential relays, any time an abrupt increase in
state magnetizing current is the core loss of the transformer excitation voltage occurs, the potential for very large
and can be estimated as: magnetizing inrush current exists. This effect is shown in
Figure 6. When the excitation voltage increases significantly,
as when the transformer is energized, this excitation voltage
causes a proportional increase in flux. The required flux
depends on several factors, such as remnant flux, point on

291
wave during switching, and transformer design. This required Emax = peak exciting voltage,
ω = system frequency,
flux can be high enough to cause the transformer core to enter
θ = voltage angle between flux and excitation voltage at time
saturation [1]. As higher values of flux are required, higher of switching
levels of magnetizing current are also required. This inrush ϕ(t) = instantaneous flux,
current can be several times the full load rating of the ϕmax = peak flux.
transformer posing a significant challenge to the differential As it was explained earlier, the instantaneous flux directly
relay. Figure 7 shows the impact inrush current has on influences the magnitude of the magnetizing inrush current
differential relaying. Fortunately, this inrush current will as shown in Figures 5 and 6. Therefore, if the flux
decay over time, sometimes taking several seconds or requirement is higher, then the magnetizing inrush current
minutes. The decay time of the inrush current is dependant on will be higher as well. Figure 8 shows the difference in flux
the R/L time constant of the system [5]. L is taken as the when the excitation voltage is applied at voltage angles of 0°
variable inductance of the transformer core and R is the and 90°.
resistive component of the system impedance and the resistive
losses of the transformer.


ϕ

H
ϕ
VDWXUDWLRQNQHH
SRQW
Time Lϕ

2α radians

Figure 6: Transformer Excitation During Saturation. Figure 8: Flux and Excitation Voltage at Full and Zero Offset.

Point on wave of switching is not typically controlled


which makes predicting the magnitude of inrush current
almost impossible.

Impact of Remnant Flux:

The point at which a transformer is deenergized influences


the amount of flux that remains in the transformer core.
Typically, the amount of flux that remains in the core is
anywhere from 30% to 80% of the maximum core flux and
can be positive or negative. The remnant flux is added to the
flux that occurs due to the excitation voltage [1]. Figure 9
shows the impact of positive and negative remnant flux on
total flux. Figure 9 assumes the excitation voltage and flux
are fully offset (90°) at the start of the event.

Figure 7: Inrush Current Impact on Differential Relaying.


ϕ

Impact of Switching Point:


ϕ

One major factor that determines the magnitude of H H


magnetizing inrush current is the point on wave at which the
Time Time
excitation voltage increases. This is because flux and exciting
voltage are linked by the following equation [1]:
6WDUWRIHYHQW 6WDUWRIHYHQW
e(t ) = Emax cos(ωt − θ ) , then the flux is defined by 3RVLWLYHUHPDQHQWIOX[ 1HJDWLYHUHPDQHQWIOX[
ϕ (t ) = ϕ max sin(ωt − θ ) + ϕ max sin(θ ) .
Where e(t) = instantaneous excitation voltage, Figure 9: Impact of Positive and Negative Remnant Flux.

292
As the point on wave the transformer is not controlled, conditions depicted in Figure 10.
neither is the point on wave at which the transformer is The amount of second harmonic component present in the
deenergized. Therefore, the amount of remnant flux present magnetizing inrush current will vary based on the saturation
in the core is an unpredictable variable. angle. Theoretically, the ratio of second harmonic to
fundamental ranges between 7% and 70% [3].
Impact of Transformer Design:

The saturation flux density and peak flux density of a


transformer core are affected by many factors, but primarily
core design [2]. The quality of electrical steel has an impact,
but has been relatively constant over the last several decades.
The major difference that has occurred in recent years has
been the increased flux density of the transformer core. As
technology has improved laminated core designs have become
more common. Laminations provide an air gap between each
lamination resulting in a lower reluctance core [1]. These
more efficient core designs lead to lower steady state
magnetization current (i.e. lower core loss) and lower
magnetization inrush currents.
As we will see in the next section this reduced excitation
current has a significant impact on transformer differential
relaying inrush detection methods.

Impact of Power System Impedance:

The strength of the source energizing the transformer


directly affects the peak magnetizing inrush current. Also, the
resistance present in the transformer windings and the source
impedance dictates how long it takes to reach steady state Figure 10: Ideal Inrush Waveform
conditions, after peak inrush. The change in flux over time is
defined by the following equation:

t +T
Δϕ = ³ (R × i )dt
t
where Δϕ = flux change per cycle,
R = total series resistance including transformer winding resistance
T = period of one cycle.

Frequency Domain Characteristics of Inrush Current:

Figure 10 shows an ideal inrush waveform created using a


real time simulator program. What is important to note about
this waveform is that it is not symmetric about the y-axis.
Any waveform that is not symmetric about the y-axis contains
significant even harmonics. Particularly, the second harmonic
tends to dominate as shown in the Fourier analysis of the
typical inrush waveform. Also note the slowly decaying DC
component of this waveform. This DC offset current is likely
to saturate the CT’s and produce some level of distortion in
the actual measured waveform [3]. An example of this
behavior is shown in Figure 11, which shows the inrush
current from a 15 MVA 66kV to 6.9kV power transformer as
captured by a protective relay. Notice the distortion
introduced in Figure 11 by CT saturation from the ideal Figure 11: Actual Inrush Waveform as captured by a protective relay

293
otherwise undesired tripping will occur. Over the years
Events Causing Inrush Current: various methods have been employed to detect and restrain
operation of the differential relay under inrush conditions.
Typically, three major causes of inrush current are This paper will discuss several of the most common
considered: transformer energization, inrush due to fault techniques utilized in electromechanical and microprocessor
clearing, and sympathetic inrush. The most severe case of relays, however, all techniques will not be discussed (for a
magnetizing inrush current results from transformer more exhaustive description see reference three).
energization. In this case there is a very large change in In electromechanical relays each phase had a separate
excitation voltage applied to the core. For three phase differential relay. There was typically a harmonic restraint
transformers, each phase will experience different peak values unit for each differential relay and restraint was applied on a
of inrush current due to the impact of the voltage angle at per phase basis. In microprocessor relays restraint no longer
time of switching. Peak current can be as much as forty times has to be applied on a per phase basis, but can be a
full load rating of the transformer bank and are typically in combination of one, two, or three phase methods.
the range of two to six times [1].
Additionally, inrush current may be caused by fault Differential Restraint Methods:
clearing activities. When a fault external to the transformer is
present the voltage applied to the transformer is depressed. One of the most common methods applied to detect
When the fault is cleared the voltage applied to the transformer inrush is based on the identification of harmonic
transformer returns to normal values. This change in voltage content. As shown in Figure 10 and 11, a significant amount
can lead to inrush currents similar to energization of the of harmonic distortion is present during an inrush event.
transformer. However, because the magnitude of the voltage However, when internal faults occur, very little second
change is not as severe, the expected inrush current will be harmonic component is present. This property has been
lower then the energization case. Also, no significant offset in utilized in various ways to restrain transformer differential
flux is expected, further reducing the chance of severe relay operation.
saturation [3]. The presence of load current will act to lower
the ratio of second harmonic to fundamental current. Method 1: Percentage of Total Harmonics
The case of sympathetic inrush occurs when two or more
transformer banks in parallel are energized sequentially. The One of the earliest methods of inrush restraint is based
first transformer bank will cause magnetizing inrush current on percentage of total harmonics. In this method, the
to flow when energized with no additional effect. However, harmonic content is compared to the fundamental current.
the energization of the second transformer can cause a very
significant voltage drop across the source resistance, see I diff ( 2 nd harmonic) + I diff ( 3 rd harmonic) + ...I diff (n harmonic)
if > setting, then block differential
Figure 12. This drop in voltage affects the previously I differential ( Fundamental )
energized transformer, causing it to draw inrush current in
the opposite direction [3]. Figure 12 also shows a typical
waveform during sympathetic inrush. This method is typically applied in electromechanical
relays on a per phase basis. The setting level is typically 20%.
This method began facing trouble as more efficient
Current

7[
transformer cores produced less magnetizing inrush current
7[ Tim
as discussed in section three of this paper.
5M; e

7[
Tim
Method 2: 2nd Harmonic Percentage of Fundamental
7[ e

In this method the second harmonic component


Current

percentage of the fundamental current is compared to a


Figure 12: Sympathetic Inrush
setting threshold:

I differential ( 2 nd harmonic)
IV. INRUSH RESTRAINT METHODS if > setting , then block differenti al
I differential ( fundamental )

In section two of this paper it was shown that magnetizing This method is widely applied in electromechanical
inrush current appears as differential current in a transformer relays on a per phase basis. This method is still popular in
differential relay. Inrush current must be properly identified microprocessor based relays, but is not restricted to simply a
and used to inhibit operation of the differential relay, per phase basis. Again, when applied on a per phase basis,

294
this method began to face problems as core designs became method dynamically lowers the restraint threshold at the
more efficient, producing less magnetizing inrush current. beginning of an inrush restraint and gradually increases back
Typical setting levels range from about 15% to 20%. to the setting value over a period of 5-6 cycles [3]. Figure 14
Electromechanical relays continuously calculate the depicts the operation of the dynamic restraint characteristic.
percentage of second harmonic, regardless of current present.
In microprocessor relays, there is typically a minimum value
of current required for the relay to recognize the signal as
valid. This can potentially cause misoperation as the
transformer inrush current decays to smaller magnitudes.
Careful consideration should be given to relay minimum
sensitivity for harmonic calculation and differential minimum
pickup to avoid problems.

Method 3: 4th Harmonic Percentage of Fundamental

In this method the fourth harmonic component percentage


of the fundamental current is compared to a setting threshold:
I differential ( 4th harmonic)
if > setting, then block differenti al
I differential ( fundamental )
Figure 13: Second Harmonic Content During Inrush [3].
Again, as seen from Figure 10 and 11 due to the lack of
symmetry about the Y axis, the waveform is rich in even R

R R
harmonics. Typically, the second harmonic is the dominant
,1+,%,7

harmonic and is present in much larger quantities then the


R R
fourth harmonic. As such, very little is gained by using the
fourth harmonic threshold in conjunction with the second
harmonic. Furthermore, because the second harmonic R SX SX 23(5$7( SX SX R

technique has been around for much longer, plenty of


empirical data regarding the percentage of second harmonic
during transformer inrush is available, making it easier for R R

the Engineer to select an appropriate setting threshold. ,1+,%,7

R R

R
Method 4: Adaptive Restraint
Figure 14: Adaptive Harmonic Restraint Characteristic
This method is applied in microprocessor relays and
utilizes not only the second harmonic magnitude, but the Method 5: Dwell Time Method
angle as well. In this method the second harmonic percentage
of fundamental threshold changes based on the angle of the Typically, this method determines an inrush condition is
second harmonic vs. fundamental component vector. present if the differential current drops close to zero for at
least ¼ of a cycle out of every cycle [3]. In a typical inrush


I2
 waveform, we expect the current to be near the steady state
I 21 = →
jω t
I1× e magnetizing current value (close to zero) during periods
→ where the core is operating in the linear region, and very high
nd
where I 2 is the 2 harmonic differential current phasor, when the core is in the saturation region. This behavior is

I 1 is the fundamental differential current phasor, shown in the ideal inrush waveform shown in Figure 10.
Ȧ is the system frequency. This method was commonly applied in solid state relays
and is not widely used in microprocessor relays. This method
In newer transformer core designs it is possible to get lower also has the disadvantage that the differential element must
ratios of second harmonic vs. fundamental current then the be delayed by one full cycle, as it takes at least this long to
typical 15% to 20% normally assumed [3]. This ratio will be make an inrush detection decision.
lower then 15% to 20% typically only on one phase and only
for a few cycles as shown in Figure 13. The adaptive restraint
method utilizes the fact that during inrush conditions the →I 21
angle is always close to 90º or 270º. The adaptive restraint

295
V. ANALYSIS OF HARMONIC RESTRAINT METHODS 250

200

150
The chosen harmonic restraint method, and whether it is
100
applied on a per phase or multi phase basis, has a great
50
impact on the dependability, security, and speed of the
0
differential relay. For purposes of this paper, security is
defined as the ability for the differential relay to not trip for -50

inrush events. Dependability is defined as the ability of the -100

differential relay to trip for all internal faults. Speed is how -150

quickly the relay will operate. Some of the above harmonic -200
-0.026 0.000 0.026 0.052 0.078 0.104

restraint methods have a minimum operate time that limits


the overall speed of the differential element. FFT Results

As with any protective relaying scheme, there is usually a 120

compromise made on either speed, dependability, or security


100
based on the specific requirements the user is trying to meet.
The methods from the previous section all have strengths and

% of Fundamental
80

weaknesses, but they will manifest as either a tendency to


60
over trip during inrush or over restrain during fault
conditions. The fundamental question a user must ask is 40

whether nuisance tripping during inrush conditions is more 20


desirable then restraining during an internal transformer
0
fault, or vice versa. This question depends on a users past 60 120 180 240 300 360 420 480 540 600 660 720

history, company practices, and operating requirements. Frequency

Impact of CT Saturation: Figure 15: Mild Case of CT Saturation [6]

As discussed earlier in the paper, the transformer inrush 600

waveform is extremely distorted when compared to a 500

fundamental frequency sinusoidal waveform, leading to many 400

harmonics in the frequency spectrum. All of the 300

aforementioned detection methods rely on the inrush 200

waveform characteristics to determine whether an inrush 100

condition is present. However, extreme distortion in the


0

measured relay current will also occur during CT saturation,


as shown in Figure 15 and 16. During high magnitude
-100

internal faults it is possible for the output of the winding CT’s -200

to have waveforms similar to Figure 15 and 16. Therefore, it -300

is impossible to set the harmonic restraint method thresholds -400


-0.026 0.000 0.026 0.052 0.078 0.104

considering only the inrush waveform.


CT saturation will impact the dependability of all methods FFT Results
discussed in this paper. For example, if using method one or
120
two it would be unwise to set the detection threshold at 5%. If
a severe internal transformer fault occurred, it would be 100

possible for the differential relay to improperly restrain


% of Fundamental

80
because the second harmonic content exceeded 5%. Notice,
that method one which uses percentage of total harmonics 60

will be less secure then the percentage of second harmonic,


40
because under CT saturation many harmonics are produced.
Due to the possibility of CT saturation, it is the authors 20

opinion that the detection threshold not be set below 15% for
0
the percent total harmonic or percent second harmonic 60 120 180 240 300 360 420 480 540 600 660 720

methods Frequency

Figure 16: Severe CT Saturation Waveform and Harmonic Spectrum [6].

296
The dwell time method is also subject to errors because of
CT saturation. As seen in Figure 16, severe CT saturation
will create flat spots in the waveform near the zero value.
This can lead to an incorrect blocking decision during high
magnitude internal faults, compromising the dependability of
the differential relay. Furthermore, the security of the dwell
time method is also compromised during normal inrush
events because of CT saturation. As shown in Figure 10, a
high DC component is present during inrush conditions,
which can cause even properly dimensioned CT's to saturate,
resulting in an offset waveform as shown in Figure 11. This
figure shows that the flat spots during CT saturation are no
longer close to zero, but are offset due to the saturation effect.
This behavior will cause the relay to incorrectly allow a
differential trip.
The adaptive restraint method security and dependability
are not impacted by CT saturation, but speed can be Figure 17: Three Phase 500 MVA, 500kV/230kV Autotransformer Inrush [1].
compromised. As previously described, the adaptive restraint
method lowers the second harmonic sensitivity threshold at
the onset of an inrush event. If the angle of I21 is close to 0 or The adaptive restraint method was specifically designed to
180 degrees inrush restraint is removed immediately [3]. deal with this problem and performs very well under these
However, in cases of mild CT saturation the angle of I21 conditions.
maybe closer to 90 or 270 degrees which can add up to six
cycles of delay. Impact Differential Inhibit Modes

Impact of Using Unrestrained Differential So far the discussion regarding dependability and security
has been limited to the per phase mode. Enhancements to the
An unrestrained differential element is often included in dependability and security of the harmonic restraint methods
microprocessor transformer differential relays. The discussed earlier can be achieved by using either one out of
unrestrained differential element operates strictly on three (cross blocking), two out of three, or averaging modes.
magnitude of the differential current and applies no restraint, Electromechanical relays are typically limited to the per
harmonic or otherwise. This element is set so that the pickup phase mode, microprocessor relays can use any combination
value exceeds the differential current expected for the most of the above modes.
severe external fault under worst case CT saturation
conditions. This element can be used to address the issue of Per Phase Blocking Mode:
improper restraining during severe internal faults because of
CT saturation. This mode makes a tripping or restraining decision on a
per phase basis. For example, if one phase makes a blocking
Impact of Modern Transformer Designs decision due to high harmonic content it will have no impact
on the other two phases. This mode will work with any of the
As transformer designs have become more efficient the detection methods discussed in section four. This method
magnetizing current has decreased. This directly impacts the adds no security or dependability against CT saturation or
amount of second harmonic present in the inrush waveform. lower magnetizing current designs.
Figure 17 is an actual autotransformer energization event that
yielded very low second harmonic currents on two phases. By 1 out of 3 Blocking Mode (Cross Blocking):
the beginning of the second cycle, the second harmonic to
fundamental ratio on two phases had decreased to 9.9% and In this mode if any one phase makes the decision to block
13.6% respectively. differential due to inrush, all phases are blocked. For
Simply lowering the second harmonic setting threshold can example, if one phase makes a blocking decision all phases
increase the security of methods one or two, but will will be blocked from operating. Often, cross blocking is
compromise dependability during severe internal faults as applied only for a short time and then removed to ensure
mentioned in the previous section. Also, the dwell time tripping occurs for a fault during energization.
method is impacted by newer transformer designs because the This mode enhances security greatly, because even in
flat periods may become shorter then ¼ cycle. newer transformer designs, at least one phase will typically

297
have a second harmonic percentage greater then 15%. This phase basis.
enhancement in security comes at some cost to dependability. If dependability and speed are the primary concerns, then
For example, assume a fault develops on one phase during the second harmonic method in the per phase or averaging
energization. On a per phase basis, this condition would be mode can be applied. Again, if speed is not a concern the
identified by the differential element on the faulted phase and adaptive method can be applied on a per phase basis.
proper tripping would occur. However, in the cross blocking If dependability, security, and speed are all primary
mode, at least one unfaulted phase will identify an inrush concerns, it is the authors opinion that the second harmonic
condition is present and block the differential relay. If a time percentage of fundamental method applied in the averaging
delayed application of cross blocking occurs, dependability is mode is the best selection.
not compromised, but speed is affected. Lastly, speed and dependability issues can potentially be
mitigated for the cross blocking and adaptive restraint modes
2 out of 3 Blocking Mode: by using an unrestrained differential element in parallel with
the restrained differential element.
With this mode, two phases must make a blocking decision
in order to prevent differential operation. For example, if only VI. IMPACT OF TRANSFORMER INRUSH ON OTHER RELAYS
one phase makes a blocking decision, then the differential
relay will still operate. A minimum of two phases must make As we have seen throughout this paper, transformer inrush
a blocking decision in order to prevent differential operation. has a significant impact on transformer differential relaying.
This mode enhances security, but not to the same degree as However, the impact of transformer inrush is not limited to
the cross blocking mode. Assuming a minimum setting of just the transformer differential relay. This section discusses
15% second harmonic, on our example transformer, there the impact of transformer inrush on various system
would not be sufficient second harmonic on two phases to protection.
block the differential relay. However, dependability is not
compromised to the same degree either. In the cross blocking Generator Differential Relays:
mode all fault types except three phase faults compromise
dependability. In the 2 out of 3 method only single phase Transformer inrush can have a significant impact on
faults can potentially compromise dependability. generator differential relaying. As seen in Figures 10 and 11
the DC component of the transformer inrush can easily cause
3 Phase Averaging Mode: the generator differential CT’s to saturate. If a generator is
electrically close to a transformer, and the transformer rating
This mode is only applicable to method one, two, and is similar to or greater then the generator rating, then care
three. This method is typically applied as shown in the must be taken to avoid misoperation of the generator
equation below: differential relay. If the generator’s supply conductors are
small enough, the differential CT’s can be connected in a flux
I A Diff 2nd + I B Diff 2nd + I C Diff 2nd balanced configuration as shown in Figure 18. In this
if > setting , then block differenti al
I A Fund 2nd + I B Fund 2nd + I C Fund 2nd configuration, no flux is placed on the CT unless a
differential condition exists eliminating the chance of
misoperation. If two separate CT’s are used as shown in
This method enhances security, again not as much as the Figure 19, then care must be taken to properly size the ’CT’s
cross blocking mode. However, dependability is not so that minimal saturation takes place. IEEE C37.110 or
compromised. If a fault occurs during energization, the reference number seven have excellent guidelines on how to
magnitude of the fundamental component should be much properly dimension CT’s. Proper CT selection is the best
higher then the second harmonic component due to the method to deal with this problem, however, incorrectly
magnetizing inrush current. designed systems may get installed. In these cases the
generator relay may have to be delayed when energizing the
Summary of Blocking Modes: transformer. This can be accomplished by using the close
signal from the transformer breaker to delay operation of the
As it can be seen from the discussion above there is no generator differential element, see Figure 20.
single right answer for how to restrain a differential relay for
transformer inrush events. If security and speed are the
primary concerns, it is advisable to utilize the second
harmonic detection method in a cross blocking or averaging
mode. If security is the primary concern and speed is not a
factor, the adaptive restraint method can be applied on a per

298
Typically, when the feeder breaker is opened the cold load
pickup scheme is armed. Identification of this condition is not
dependant on any other method, including those discussed in
this paper.

Line Differential Relaying:

Unless there is a transformer within the protected zone,


transformer inrush generally does not affect distance or line
87G Relay current differential relaying.

Figure 18: Flux Balanced CT Configuration. VII. SUMMARY

Transformer inrush occurs anytime the excitation voltage


applied to a transformer is changed. Inrush current
magnitude is typically anywhere between 2 and 8 times the
transformer full load rating [5]. Inrush current magnitude and
duration is dependant upon the point at which the transformer
excitation voltage is applied, the remnant flux in the core,
and the resistive component of the power system and
transformer impedance. Furthermore, transformer inrush can
occur due to transformer energization, energization of a
parallel transformer bank, and post fault voltage recovery.
87G Relay Transformer inrush must be properly accounted for when
setting transformer differential relays, otherwise improper
Figure 19: Separate Line and Neutral CT Configuration. tripping can occur. Various methods have been devised to
properly identify inrush conditions and maintain differential
relay security, speed, and dependability. The effectiveness of
detection methods is impacted by CT saturation, increasingly
efficient transformer designs, and nature of the detection
87G method. For second harmonic percentage methods, the author
recommends that a setting threshold above 15% should be
Transformer Close CB Command
Delays 87G Relay used to avoid compromising dependability under severe CT
saturation. No single method is perfect and the individual
user requirements will dictate the ultimate selection.
Figure 20: Preventing Generator Differential Misoperation Transformer inrush can impact power system relaying
beyond the transformer differential relay. Particularly,
Motor Protection Relays:
generator differential and feeder protection relays can be
significantly impacted by transformer inrush.
In a typical application transformer magnetizing inrush has
Finally, regardless of which method is chosen careful
no impact on motor protection relays.
attention should be given to event and waveform capture
functionality. In today’s world of microprocessor relays it is
Feeder Protection Relays:
very easy to obtain waveform and event data for
misoperations and correct operations. Because no method is
On distribution feeders after loss of voltage, it is possible
perfect it behooves the user to collect inrush and fault
that many small distribution transformers can go into inrush
waveforms whenever possible to refine setting thresholds.
simultaneously upon closing the feeder breaker. This effect is
This data can also be used to evaluate whether the proper
commonly known as cold load pickup and can cause
cross blocking mode is selected.
instantaneous elements or short time delay time overcurrent
relays to misoperate. Most modern microprocessor relays
have a cold load pickup function that can temporarily
increase settings during this condition. Cold load pickup is
typically identified by tracking status of the feeder breaker

299
REFERENCES
[1] Rich Hunt, Joe Schaefer, Bob Bentert: "Practical Experience in Setting
Transformer Differential Inrush Restraint" Proceedings of the Georgia
Institute of Technology Protective Relaying Conference. Atlanta, GA May
2007.
[2] J. Berdy, W. Kaufman, K. Winick: “A Dissertation on Power Transformer
Excitation And Inrush Characteristics”. Proceedings of the Georgia
Institute of Technology Protective Relaying Conference. Atlanta, GA May
1976.
[3] Bogdan Kasztenny , Ara Kulidgian: “An Improved Transformer Inrush
Restraint Algorithm Increases Security While Maintaining Fault Response
Performance.” Proceedings of the Texas A&M Annual Conference for
Protective Relay Engineers. College Station, TX. April 2004.
[4] Elmore W.A., “Applied Protective Relaying”. Westinghouse Electric
Corporation, 1982.
[5] “Electrical Transmission and Distribution Reference Book.” ABB Electric
Systems Technology Institute. Raleigh, North Carolina, 1997
[6] “IEEE PSRC CT Saturation Calculator” IEEE Power Engineering Society
Protective Relaying Subcommittee, www.pes-psrc.org
[7] Rich Hunt, “Impact of CT Errors on Protective Relays – Case Studies and
Analysis.” Proceedings of the Georgia Institute of Technology Protective
Relaying Conference. Atlanta, GA May 19-20, 2008.

BIOGRAPHICAL SKETCHES

Suhag Patel is an Application Engineer in the area of Protection & Automation


systems for General Electric. He received a BSEE degree from the University of
California, Los Angeles (UCLA) and an MSEE with an emphasis in Power
Systems from California State University, Long Beach (CSULB). Mr. Patel is a
Registered Professional Engineer in the state of California and an active member
of the IEEE Power Systems Relay Committee. Prior to his work at General
Electric, Mr. Patel was a Regional Technical Manager for Asea Brown Boveri
(ABB) and held various engineering positions at Southern California Edison and
Shell Oil Products.

ACKNOWLEDGEMENT
This paper builds heavily on the work done by Rich Hunt of GE-Multilin from
past papers. His help was invaluable in creating this document.

300

Das könnte Ihnen auch gefallen